European Journal of Social Psychology Eur. J. Soc. Psychol. 28, 475±481 (1998) Minimal conditions for the creation of a unit relationship: the social bond between birthdaymates DALE T. MILLER*, JULIE S. DOWNS and DEBORAH A. PRENTICE Department of Psychology, Princeton University, U.S.A. Abstract We hypothesize that sharing a birthday is su cient to create a unit relationship. Two studies demonstrated that individuals cooperated more in a prisoners dilemma game when their ( ctitious) opponent shared their birthday. They also reacted more negatively to betrayal and were less sensitive to relative gains for self versus other. # 1998 John Wiley & Sons, Ltd. INTRODUCTION The idea that objects or individuals sharing an attribute tend to be grouped and perceived as a unit is one of the oldest in psychology. Early associationist theories (Robinson, 1932; Shepard & Fogelsonger, 1913) and gestalt theories (Ko ka, 1935; Kohler, 1929) both emphasized the powerful cognitive-perceptual tendency to group objects or people by similarity. Ko ka (1935), an in uential gestalt theorist spoke of there being a `... stress towards group formation which arises from mere similarity' (p. 661), and Fritz Heider (1958), the social psychologist most in uenced by the gestalt and associationist traditions, coined the term unit relation to refer to the connection that exists between people who are `perceived as belonging together in a speci cally close way' (p. 201). The present research attempts to demonstrate just how minimal a degree of similarity between two people is necessary for them to have a sense of `belonging together'. The source of similarity we chose was a shared birthday. The present research used the Prisoner's Dilemma paradigm (Rapport & Chammah, *Address for correspondence: Dale Miller, Department of Psychology, Princeton University, Green Hall, Princeton University, Princeton, NJ 08544-1010, U.S.A. Contract grant sponsor: National Institute of Mental Health Contract grant number: MH 44069 CCC 0046±2772/98/030475±07$17.50 Received 4 February 1997 # 1998 John Wiley & Sons, Ltd. Accepted 20 August 1997
476 D. T. Miller et al. 1965) to examine the hypothesis that sharing a birthdayða seemingly super cial and non-diagnostic commonalityðis su cient to create a social bond between people (Brown, Novick, Lord, & Richards, 1992; Finch & Cialdini, 1989). EXPERIMENT 1 Our rst experiment tested the hypothesis that individuals would play more cooperatively in the PD when their opponent was alleged to share their birthday than when their opponent was alleged to have a di erent birthday. We also expected the bond created by a shared birthday to lead players to react more negatively to violations of their trust by their opponent. Method Participants Sixty-four female undergraduates were recruited by telephone to participate in a `game playing experiment'. Their birthdates were identi ed prior to their participation from a university publication. Design and Procedure Although participants were run individually, they were led to believe that they were interacting with another participant. Before they began playing the game, they were asked to enter some identifying information (including their birthday) that was to be sent to the other player in an adjacent room. As they started to enter their information, the computer interrupted them with the ( ctitious) other player's information. This information revealed the other player to have the same birthday as the participant's (birthday condition), or a birthday approximately 4 months earlier or 4 months later in the same year (control condition). Participants then nished entering their own information. The game consisted of two rounds of 24 successive one-trial prisoner's dilemma games. On each trial, participants chose either action `A' (the cooperative response) or action `B' (the competitive response) and received no information about which action the other player chose. The exact values in the payo matrix varied from trial to trial, and the matrices appeared in random order. At the end of the rst round, participants indicated their assumptions about the other player's actions and general similarity to them, and then received the alleged results from the rst round. The bogus feedback indicated that the other player had chosen action `A' either ve more times than the participant (cooperative opponent) or ve fewer times than the participant (exploitative opponent). 1 Participants then played the second round of the game. 1 The cooperation rates of ve participants were either too high or too low to statistically accommodate this manipulation. The feedback to these participants simply indicated that their opponent had cooperated the
Social bond between birthdaymates 477 Results and Discussion First-Round Behaviour An analysis of variance (ANOVA) on rst-round cooperation rates revealed that, as hypothesized, birthday participants cooperated more often (M ˆ 52 per cent of trials) than did control participants (M ˆ 38 per cent of trials), F(1,58) ˆ 5.64, p 5 0.05. This di erence was not due to di erential perceptions of similarity, as birthday and control participants rated their opponents as equally similar to themselves (Ms ˆ 4.90 and 4.53, respectively, on a 1±7 scale. F(1,58) ˆ 1.94, p 4 0.10). Further, an analysis of covariance (ANCOVA) controlling for perceived similarity still revealed higher cooperation rates in the birthday condition (M ˆ 51 per cent, adjusted) than in the control condition (M ˆ 38 per cent, adjusted), F(1,57) ˆ 4.63, p 5 0.05. Second-Round Behaviour To examine the impact of the feedback about the other player's actions provided after the rst round, we conducted a 2 (bond: birthday versus control) 2 (opponent's behaviour: cooperative versus exploitive) ANCOVA on cooperation rates in round 2, controlling for cooperation rates in round 1. A main e ect of opponent's behaviour emerged, F(1,55) ˆ 10.83, p 5 0.002, indicating that participants who discovered that their opponents had played cooperatively in round 1 cooperated more in round 2 (M ˆ 47 per cent, adjusted) than did participants who discovered that their opponents had played exploitatively (M ˆ 35 per cent, adjusted). More importantly, the predicted bond opponent's behaviour interaction was also signi cant, F(1,55) ˆ 5.07, p 5 0.05. Learning of one's opponent's per dy suppressed cooperation much more in the birthday condition (M ˆ 29 per cent, adjusted) than in the control condition (M ˆ 42 per cent, adjusted). Interestingly, the reverse did not hold for a cooperative opponent: learning of one's opponent's greater cooperativeness did not have a greater impact on the behaviour of birthday participants (Ms ˆ 49 per cent, adjusted) than of control participants (M ˆ 45 per cent, adjusted). This asymmetry is consistent with previous research showing that, in general, competition has a much greater e ect than cooperation on dyad behaviour in the Prisoner's Dilemma game (e.g. Pines, 1976). The present ndings suggest that this greater sensitivity to competition was especially acute among participants who believed their opponent shared their birthday. In an ANCOVA controlling for expectations, perceived similarity, and level of cooperation in the rst round, both the main e ect of opponent's behaviour, F(1,54) ˆ 10.69, p 5 0.002, and the bond opponent's behaviour interaction, F(1,54) ˆ 4.91, p 5 0.05, remained signi cant. maximum (100 per cent) or the minimum (0 per cent), respectively. Speci cally, the two participants in the cooperative opponent condition whose cooperation rates were over 80 per cent were told that their opponent had cooperated 100 per cent of the time, and the three participants in the exploitative partner condition whose cooperation rates were under 20 per cent were told that their opponent cooperated 0 per cent of the time. This modi cation meant, e ectively, that the feedback manipulation was weakened for ve of the participants.
478 D. T. Miller et al. EXPERIMENT 2 Experiment 2 sought to demonstrate that the social bond created by a shared birthday leads people not only to demand more of their partners but also to sacri ce more for them. Speci cally, we predicted that the relative gain for self versus other built into the payo should a ect the behaviour of players less when they share a birthday than when they do not. Method Participants Participants were 49 (30 male and 19 female) undergraduates recruited by a sign-up sheet to participate in a `game playing experiment'. Design and Procedure The experiment employed a 2 (bond: birthday versus control) 3 (matrix type: favourable-to-self versus equal payo versus favourable-to-other) design. The bond manipulation and general procedure were identical to those employed in Experiment 1. The game consisted of 28 trials of a PD-like experimental game. To vary the degree of personal versus joint gain, we devised three di erent types of payo matrices. In equal payo matrices, both players earned the same number of points if they both cooperated (e.g. self ˆ 55, other ˆ 55); in favourable-to-self matrices, the participant earned more points than the other players when both cooperated (e.g. self ˆ 60, other ˆ 50); and in favourable-to-other matrices, the other player earned more points than the participant when both cooperated (e.g. self ˆ 50, other ˆ 60). The game included 12 equal payo matrices, eight favourable-to-self matrices, and eight favourable-to-other matrices. Participants received no feedback about their opponent's behaviour during the game. Results and Discussion A 2 (bond: birthday versus control) 3 (matrix type: favourable-to-self versus equal versus favourable-to-other) ANOVA on cooperation rates revealed a signi cant main e ect of bond, F(1,43) ˆ 5.52, p 5 0.05. Participants who believed they shared a birthday with the other player cooperated more than did control participants across favourable-to-self (Ms ˆ 68 per cent versus 53 per cent), equal (Ms ˆ 60 per cent versus 44 per cent), and favourable-to-other (Ms ˆ 62 per cent versus 34 per cent) matrices. The main e ect of bond remained signi cant in an ANCOVA, controlling for participants' expectations of the other player's actions, F(1,42) ˆ 4.50, p 5 0.05. The ANOVA also revealed a signi cant e ect of matrix type, F(2,86) ˆ 10.61, p 5 0.0001, and a marginally signi cant bond matrix type interaction, F(2,86) ˆ 3.05, p ˆ 0.055. (For the focused comparison between the favourable-to-self and
Social bond between birthdaymates 479 Figure 1. Mean cooperation rates by type of bond and favourability of payo matrix to self versus other favourable-to-other matrices, the predicted bond matrix type interaction was signi cant, F(1,43) ˆ 6.56, p 5 0.05). As Figure 1 shows, the prospect of a greater relative gain for one's opponent than for oneself suppressed cooperation among control participants more than among birthday participants. In an ANCOVA controlling for expectations, both the main e ect of matrix, F(1,42) ˆ 4.57, p 5 0.05, and the interaction, F(1,42) ˆ 7.38, p 5 0.01, remained signi cant. GENERAL DISCUSSION The present studies suggest that a social unit can be formed between two people sharing nothing more than a common birthday. The question of why merely sharing a birthday is su cient to induce people to act as though they `belong together in a speci cally close way' remains open. Some accounts can be ruled out, however. For example, we can state with some certainty that the feelings of closeness that arise between birthdaymates do so in the absence of any strong assumptions of general similarity. We found no di erences in perceived similarity between birthday and control participants, and the e ects of a shared birthday on levels of cooperation remained even when perceptions of similarity were covaried out. Moreover, the interaction e ects found in the present study cannot be explained by positing di erences in perceived similarity. The perception that another shared their birthday led participants not only to increase their general level of cooperation but also to react
480 D. T. Miller et al. more negatively to attempted exploitation by the other and more positively to the opportunity to bene t the other at the expense of the self. Perceived similarity alone does not seem su cient to account for these qualitative di erences between the two types of relationships. Similarity may a ect behaviour in both unit and non-unit relationships but it alone does not transform the former into the latter. The unit relationship formed between people sharing a birthday may derive solely from the sharing of a special and distinctive attribute of the self (Turnbull, Miller, & MacFarland, 1990). Our birthday, although not highly diagnostic of the kind of person we are, is both special and unique to us. As such, we feel an attachment to it and possibly, by extension, to those who share it with us. 2 Should we conclude, then, that a shared birthday constitutes an important or a trivial source of similarity? The answer to this question would seem to depend on which of its consequences we take as indicative of its importance. On the one hand, a shared birthday is important enough to lead individuals to react to each other as though they share a unit relationship. On the other hand, a shared birthday is not even important enough to lead individuals to infer that they share additional traits in common. In short, the attachment that one feel's to one's birthdaymates typi es a kind of relation that is not wellexplained by traditional notions of attribute similarity and shared category membership. The nature of this attachment and its role in interpersonal relations deserve further investigation. ACKNOWLEDGEMENTS This research was supported by Grant MH44069 from the National Institute of Mental Health. REFERENCES Brown, J. D., Novick, N. J., Lord, K. A., & Richards, J. M. (1992). When Gulliver travels: Social context, psychological closeness, and self-appraisals. Journal of Personality and Social Psychology, 62, 717±727. Finch, J. F., & Cialdini, R. B. (1989). Another indirect tactic of (self-) image management. Boosting. Personality and Social Psychology Bulletin, 15, 222±232. Heider, F. (1958). The psychology of interpersonal relations. Hillsdale, NJ: Lawrence Erlbaum Associates. Ko ka, K. (1935). Principles of gestalt psychology, New York: Harcourt Brace. Kohler, W. (1929). Gestalt psychology. New York: Liveright. Pines, A. (1976). The shift to competition in the Prisoner's Dilemma game. Social Behavior and Personality, 4, 177±186. 2 Other special and distinctive but non-diagnostic possessions of ours may function similarly. An example of such a possession is the name of our home town. People typically feel a bond with those who reside in their home town but they also may feel a bond with those who reside in other towns with the same name. Such a bond might explain the altruism shown toward the residents of Quincy, Illinois (U.S.A.) by the residents of Quincy, Massachusetts (U.S.A.) during the summer of 1993. That summer, the Mississippi river ooded for several weeks, devastating towns on either side of its banks. When residents of Quincy, Massachusetts discovered that Quincy, Illinois was one of those towns, they undertook extensive e orts to send relief supplies to the residents of their town's namesake. The bond created by sharing the name Quincy apparently was su cient to motivate their actions of altruism.
Social bond between birthdaymates 481 Rapoport, A., & Chammah, A. M. (1965). Prisoner's dilemma. Ann Arbor: University of Michigan. Robinson, E. S. (1932). Association theory today. New York: Liveright. Shepard, G. F., & Fogelsonger, H. M. (1913). Studies in association and inhibition. Psychological Review, 20, 290±311. Turnbull, W., Miller, D. T., & McFarland, C. (1990). Distinctiveness, identity, and bonding. In J. M. Olson and M. P. Zanna (Eds), Processes in self-perception: The Ontario Symposium (Vol. 6, pp. 115±133), Hillsdale, NJ: Erlbaum.